Hands-On Convolutional Neural Networks with TensorFlow
ConvolutionalNeuralNetworks(CNN)areoneofthemostpopulararchitecturesusedincomputervisionapps.ThisbookisanintroductiontoCNNsthroughsolvingreal-worldproblemsindeeplearningwhileteachingyoutheirimplementationinpopularPythonlibrary-TensorFlow.Bytheendofthebook,youwillbetrainingCNNsinnotime!Westartwithanoverviewofpopularmachinelearninganddeeplearningmodels,andthengetyousetupwithaTensorFlowdevelopmentenvironment.Thisenvironmentisthebasisforimplementingandtrainingdeeplearningmodelsinlaterchapters.Then,youwilluseConvolutionalNeuralNetworkstoworkonproblemssuchasimageclassification,objectdetection,andsemanticsegmentation.Afterthat,youwillusetransferlearningtoseehowthesemodelscansolveotherdeeplearningproblems.Youwillalsogetatasteofimplementinggenerativemodelssuchasautoencodersandgenerativeadversarialnetworks.Lateron,youwillseeusefultipsonmachinelearningbestpracticesandtroubleshooting.Finally,youwilllearnhowtoapplyyourmodelsonlargedatasetsofmillionsofimages.
·5.2萬字